Traffic Identification of Mobile Apps Based on Variational Autoencoder Network

Ding, Li; Zhu, Yuefei; Lin, Weihong

doi:10.1109/cis.2017.00069

Cited by 24 publications

(13 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, deep learning has also been applied for mobile app identification. In [47], a two-stage learning was proposed that performs unsupervised feature extraction in the first stage. The first stage does not need labeled data and can use public unlabeled data to improve accuracy.…”

Section: B Smartphone Traffic Identificationmentioning

confidence: 99%

Large-Scale Mobile App Identification Using Deep Learning

2020

View full text Add to dashboard Cite

Many network services and tools (e.g. network monitors, malware-detection systems, routing and billing policy enforcement modules in ISPs) depend on identifying the type of traffic that passes through the network. With the widespread use of mobile devices, the vast diversity of mobile apps, and the massive adoption of encryption protocols (such as TLS), large-scale encrypted traffic classification becomes increasingly difficult. In this paper, we propose a deep learning model for mobile app identification that works even with encrypted traffic. The proposed model only needs the payload of the first few packets for classification, and, hence, it is suitable even for applications that rely on early prediction, such as routing and QoS provisioning. The deep model achieves between 84% to 98% accuracy for the identification of 80 popular apps. We also perform occlusion analysis to bring insight into what data is leaked from SSL/TLS protocol that allows accurate app identification. Moreover, our traffic analysis shows that many apps generate not only app-specific traffic, but also numerous ambiguous flows. Ambiguous flows are flows generated by common functionality modules, such as advertisement and traffic analytics. Because such flows are common among many different apps, identifying the source app that generates ambiguous flows is challenging. To address this challenge, we propose a CNN+LSTM model that uses adjacent flows to learn the order and pattern of multiple flows, to better identify the app that generates them. We show that such flow association considerably improves the accuracy, particularly for ambiguous flows. Furthermore, we show that our approach is robust to mixed traffic scenarios where some unrelated flows may appear in adjacent flows. To the best of our knowledge, this is the first work that identifies the source app for ambiguous flows.

show abstract

Section: B Smartphone Traffic Identificationmentioning

confidence: 99%

Large-Scale Mobile App Identification Using Deep Learning

2020

View full text Add to dashboard Cite

show abstract

“…On the contrary, unlabeled traffic data is abundant and readily available. Therefore, some researchers began to explore how to use easily-obtainable unlabeled traffic data combined with a few labeled traffic data for accurate traffic classification [27], [31]. Actually, this is a typical semisupervised learning, by which one can pre-train a model D u with a large unlabeled traffic data and then transfer it to a new architecture and retrain the model with D l .…”

Section: E Pre-training Designmentioning

confidence: 99%

“…Besides, large dataset will consume enormous computing and memory resources. In [20], [21], they designed a SAE model used in the pre-training process for dimension reduction, while VAE was used in [27], [31] for semi-supervised learning to overcome the labeling problem. A summary of pre-training desigh of existing work is shown in Table . 4.…”

Section: E Pre-training Designmentioning

confidence: 99%

See 1 more Smart Citation

A Survey of Techniques for Mobile Service Encrypted Traffic Classification Using Deep Learning

et al. 2019

View full text Add to dashboard Cite

The rapid adoption of mobile devices has dramatically changed the access to various networking services and led to the explosion of mobile service traffic. Mobile service traffic classification has been a crucial task that attracts strong interest in mobile network management and security as well as machine learning communities for past decades. However, with more and more adoptions of encryption over mobile services, it brings a lot of challenges about mobile traffic classification. Although classical machine learning approaches can solve many issues that port and payload-based methods cannot solve, it still has some limitations, such as time-consuming, costly handcrafted features, and frequent features update. With the excellent ability of automatic feature learning, Deep Learning (DL) undoubtedly becomes a highly desirable approach for mobile services traffic classification, especially encrypted traffic. This survey paper looks at emerging research into the application of DL methods to encrypted traffic classification of mobile services and presents a general framework of DL-based mobile encrypted traffic classification. Moreover, we review most of the recent existing work according to dataset selection, model input design, and model architecture. Furthermore, we propose some noteworthy issues and challenges about DL-based mobile services traffic classification.

show abstract

“…To overcome this issue, other works propose classifiers based on deep learning, that work directly on input data by automatically distilling structured and complex feature representations at the expense of a higher training complexity and need for larger datasets [14]. In wireless networks, this approach has been considered via variational autoencoder networks [21], convolutional networks [22] or multi-modal classifiers [6] [23]. Nonetheless, as explained above, using SL flow-based classifiers in mobile networks requires a large training dataset and implies installing probes in the core network, which is undesirable for network operators.…”

Section: Related Workmentioning

confidence: 99%

Encrypted Traffic Classification Based on Unsupervised Learning in Cellular Radio Access Networks

et al. 2020

View full text Add to dashboard Cite

Traffic classification will be a key aspect in the operation of future 5G cellular networks, where services of very different nature will coexist. Unfortunately, data encryption makes this task very difficult. To overcome this issue, flow-based schemes have been proposed based on payload-independent features extracted from the Internet Protocol (IP) traffic flow. However, such an approach relies on the use of expensive traffic probes in the core network. Alternatively, in this paper, an offline method for encrypted traffic classification in the radio interface is presented. The method divides connections per service class by analyzing only features in radio connection traces collected by base stations. For this purpose, it relies on unsupervised learning, namely agglomerative hierarchical clustering. Thus, it can be applied in the absence of labeled data (seldom available in operational cellular networks). Likewise, it can also identify new services launched in the network. Method assessment is performed over a real trace dataset taken from a live Long Term Evolution (LTE) network. Results show that traffic shares per application class estimated by the proposed method are similar to those provided by a vendor report. INDEX TERMS Traffic classification, radio access network, trace, unsupervised learning, clustering.

show abstract

Traffic Identification of Mobile Apps Based on Variational Autoencoder Network

Cited by 24 publications

References 3 publications

Large-Scale Mobile App Identification Using Deep Learning

Large-Scale Mobile App Identification Using Deep Learning

A Survey of Techniques for Mobile Service Encrypted Traffic Classification Using Deep Learning

Encrypted Traffic Classification Based on Unsupervised Learning in Cellular Radio Access Networks

Contact Info

Product

Resources

About